Application of a YOLOv8-Based Monocular Ground-Plane Intersection Method to Pothole Detection

Targeting the problem of road pothole detection and distance estimation under controlled experimental conditions, this paper proposes a method for pothole identification and ranging based on monocular vision and the ground-plane intersection method. The camera is calibrated using Zhang Zhengyou’s checkerboard calibration method to obtain its intrinsic and extrinsic parameters along with distortion coefficients. This is combined with the YOLOv8 object detection model to achieve pothole identification. Furthermore, a pothole distance estimation model based on the ground-plane intersection method is constructed. The model is validated through sandbox experiments conducted on a smart-car platform equipped with a Raspberry Pi 5B. Experimental results demonstrate the preliminary feasibility of the proposed method for short-range pothole detection and distance estimation in a controlled sandbox environment, while further validation under more complex real-road conditions is still needed.